Docker Introduction and its Usage in Machine Learning

Docker: Introduction and its
Usage in Machine Learning
By: Yogendra Singh
IIT Indore

• Lightweight, open, secure platform
• Simplify building, shipping, running apps
• Runs natively on Linux or Windows
Server
• Runs on Windows or Mac
Development machines (with a
virtual machine)
• Relies on "images" and "containers"
Docker is a software platform that allows you to build, test, and
deploy applications quickly, packaging software into
standardized units called containers.

Get Docker
• Installation instructions available in https://docs.docker.com/get-docker/
• For Linux, Windows and Mac OS

Session Lab Details
• Host: atal.iiti.ac.in
• Username: atuser1, atuser2, atuser3……atuser10
• Password: atal@1234

What can I do…..“Its running on my PC”!!!!

What are containers?
Source: docs.docker.com
• Container ≠ VM
• Isolated
• Share OS
• and sometimes bins/libs

• Standardized packaging for
software and dependencies
• Isolate apps from each other
• Share the same OS kernel
• Works for all major Linux
distributions
• Containers native to Windows
Server 2016
…..What is a container?

Docker Image
Example: Ubuntu with Node.js and
Application Code
Docker Container
Created by using an image. Runs
your application.
The Role of Images and Containers

Some Docker vocabulary
Docker Image
The basis of a Docker container. Represents a full application
Docker Container
The standard unit in which the application service resides and executes
Docker Engine
Creates, ships and runs Docker containers deployable on a physical or virtual, host
locally, in a datacenter or cloud service provider
Registry Service (Docker Hub (Public) or Docker Trusted Registry (Private)
Cloud or server based storage and distribution service for your images

Basic Docker Commands
$ docker pull node:latest
$ docker ps
$ docker container run –d –p 5000:5000 –-name node node:latest
$ docker ps
$ docker stop node(or <container id>)
$ docker rm node (or <container id>)
$ docker rmi (or <image id>)
$ docker --help

Basic Docker Commands
$ docker logs <cid>
$ docker run -it -p 8888:8888 -v /home/atal/www/:/home/jovyan/ jupyter/tensorflow-
notebook:latest
$ docker inspect
$ docker ps
$ docker stop node(or <container id>)
$ docker rm node (or <container id>)
$ docker rmi (or <image id>)
$ docker --help

14
Dockerfile – Linux Example
• Instructions on
how to build a
Docker image
• Looks very similar
to “native”
commands
• Important to
optimize your
Dockerfile

Section 2:
Anatomy of a Docker Container
Docker Volumes
Volume Use Cases

Let’s Go Back to Our Dockerfile
17

Each Dockerfile Command Creates a Layer
Kernel
FROM
18
RUN
WORKDIR
COPY
EXPOSE
…

Docker Image Pull: Pulls Layers
19

Docker Volumes
20
• Volumes mount a directory on the host into the container at a specific location
• Can be used to share (and persist) data between containers
• Directory persists after the container is deleted
• Unless you explicitly delete it
• Can be created in a Dockerfile or via CLI

Why Use Volumes
21
• Mount local source code into a running container
docker container run -v $(pwd):/usr/src/app/
myapp
• Improve performance
− As directory structures get complicated traversing the tree can slow system
performance
• Data persistence

What is Docker Bridge Networking
Docker host
bridgenet1
Cntnr 1 Cntnr 2 Cntnr 3
Docker host
bridgenet2
Cntnr 4 Cntnr 5 Cntnr 6
bridgenet3
Cntnr 7
docker network create -d bridge --name bridgenet1
23

Docker Bridge Networking and Port Mapping
Docker host 1
Bridge
Cntnr1
10.0.0.8
L2/L3 physical network
:80
:8080
172.14.3.55
$ docker container run -p 8080:80 ...
Host port
24
Container port

Docker Compose: Multi Container Applications
49
• Build and run one container at a time
• Manually connect containers together
• Must be careful with dependencies and start
up order
• Define multi container app in compose.yml file
• Single command to deploy entire app
• Handles container dependencies
• Works with Docker Swarm, Networking,
Volumes, Universal Control Plane

version: '2' # specify docker-compose version
# Define the services/containers to be run
services:
angular: # name of the first service
build: client # specify the directory of the Dockerfile
ports:
- "4200:4200" # specify port forewarding
express: #name of the second service
build: api # specify the directory of the Dockerfile
ports:
- "3977:3977" #specify ports forewarding
database: # name of the third service
image: mongo # specify image to build container from
ports:
- "27017:27017" # specify port forewarding
Docker Compose: Multi Container Applications

Docker images
• Docker Hub
• Dockerfile

Dockerfile
• It is possible to build your own images reading instructions
from a Dockerfile
FROM centos:7
RUN yum install -y python-devel python-virtualenv
RUN virtualenv /opt/indico/venv
RUN pip install indico
COPY entrypoint.sh /opt/indico/entrypoint.sh
EXPOSE 8000
ENTRYPOINT /opt/indico/entrypoint.sh

docker-compose
• Allows to run multi-container Docker applications reading
instructions from a docker-compose.yml file
version: "2"
services:
my-application:
build: ./
ports:
- "8000:8000"
environment:
- CONFIG_FILE
db:
image: postgres
redis:
image: redis
command: redis-server --save "" --appendonly no
ports:
- "6379"

Docker Introduction and its Usage in Machine Learning

Recommended

More Related Content

Similar to Docker Introduction and its Usage in Machine Learning (20)

Recently uploaded (20)

Docker Introduction and its Usage in Machine Learning